"In Shell’s Bellaire lab, one of the nation’s best-funded research facilities, sat the contraption Willis had assembled at home over the weekend. It was a goldfish bowl, filled with liquid Knox gelatin and some plaster in it. Willis had used the gelatin to simulate rock—appropriate, given Hubbert’s work on laws of scaling—and had stuck an Alka-Seltzer bottle in the middle of it to mimic a well. He’d put the liquid gelatin in the fridge and let it set, then pulled out the bottle. Then he’d used a baster to pump a slurry of plaster of Paris down the hole, filling it until the plaster began to push its way into the gelatin, forming fractures. As their theory predicted, the fractures were vertical. Although Willis’s setup was kludged together, Hubbert immediately realized it was what they needed to win over the field engineers: a clear demonstration. They’d have an opportunity to make their case at an internal Shell conference in early 1956, in several weeks’ time. They got to work on building a larger version of the model. To replace Willis’s goldfish bowl, Hubbert scoped out bigger aquariums on sale at local shops. At the Shell conference, Hubbert and Willis explained their experiment and showed the plaster casts, first from one angle, with the fractures flaring out from either side of the well. Then they rotated the cast, so the audience could see that the fractures were thin and sharp, like a knife’s blade. And of course, they were undoubtedly vertical. Within a week of this demonstration, field engineers began sending in data they’d collected after fracturing wells. Some of them had put rubber plugs down wells to form an impression of the wall. Others sent cameras down the hole. This field data showed the fractures were indeed vertical. The theory was right—and finally the engineers believed it. Willis’s contraption “had a magical effect,” as Hubbert put it. “It made Christians out of these people.”"